Salt Security identifies agentic AI security risks in autonomous API workflows

The rapid deployment of autonomous agents into production environments creates a critical vulnerability for streaming platforms relying on automated content moderation and metadata management. Because these agents maintain persistent memory and execute multi-step API workflows, a single successful injection can lead to long-term data poisoning or unauthorized system access. For the streaming ecosystem, this necessitates a shift from simple model filtering to granular identity management for non-human actors. As platforms integrate more machine-to-machine traffic to handle massive content libraries, the security of the underlying API layer becomes the primary bottleneck for safe automation. Watch for streaming providers to adopt NIST or OWASP frameworks specifically to audit autonomous agent decision-making logs.

Nokia and Ericsson have both moved agentic AI from pilot programs into production network environments during 2026, establishing the competitive landscape that makes securing autonomous agent workflows an urgent priority. In June 2026, Ericsson launched its AI in RAN commercial software subscription claiming up to 20% higher downlink throughput across more than 15 live deployments using existing baseband silicon. Nokia followed with its own agentic AI framework embedded in the Network Services Platform, letting carriers deploy AI agents that make decisions from real-time network data inside operator-defined guardrails, with commercial availability expected by end of 2026. Verizon disclosed that its 60,000-site vRAN now applies agentic AI to configuration changes and service assurance, while publicly calling for industry-wide interoperability standards for agentic systems.

The business architecture around these agentic deployments is consolidating rapidly, raising the stakes for security teams managing machine-to-machine API traffic. Nokia announced partnerships with AWS and Databricks to build a unified data, cloud, and control layer for autonomous networks, positioning its Autonomous Network Fabric as an operating system spanning radio, core, transport, and service domains. The Databricks integration addresses fragmented telco data silos with code-once workflows designed to reduce platform lock-in, while the AWS deployment brings cloud scalability to orchestration, assurance, and inventory management. Nokia reported that operators using its autonomous networks portfolio are already achieving automation rates above 90 percent and service delivery times of four hours or less. These multi-vendor, multi-cloud agentic architectures multiply the API surface area that security platforms like Salt's must monitor.

The technical divergence between Ericsson and Nokia on AI-RAN architecture has direct implications for how agentic AI systems interact with underlying infrastructure. Light Reading reported that Nokia is designing its entire Layer 1 RAN to run on Nvidia GPUs via CUDA, while Ericsson limits GPU use to the FEC function only, meaning the two vendors expose fundamentally different compute surfaces to autonomous agents. , highlighting the split between incremental RAN intelligence and broader shared AI compute models. This architectural fragmentation means that agentic AI security controls cannot assume a uniform execution environment, reinforcing Salt Security's emphasis on granular identity management and continuous monitoring of machine-to-machine API traffic across heterogeneous deployments.